Testing

[alsonicr]

Here you can write all the potential changes we should provide before the next version is released.

[alecristia]

outward facing things

it may be easier for users if functions are labeled homogeneously

so assuming find.rating.segment stays the same, then:

• aggregate.rating OK
• reliability --> get.reliability
• SDT.raterData --> get.classification
• raterComparaison--> compare.rating

intermediate things (that users see as printouts and/or in the help)

• Record span -- in aggregate_rating.R -- not sure what this means; do we mean how long the recordings span? or rather how long the annotations span? if the latter, let's change this to: "Annotations span"
• length of recording annotation -- in aggregate_rating.R -- how is different from the previous one?
• harmonize how the class is referred to, eg ChildRecordings : a ChildRecordings class -- should it be ChildRecordingsData class?
• error message "is not a ChildRecordings class retrun null result" is not super clear -- perhaps it's missing a word: "is not a ChildRecordings class because it returned a null result"; or just say: "is not a ChildRecordings class"

• in find_rating_segment.R:

  • #' @param filename : a wav file name to consider -- or also a list of file names?
  • I don't understand this phrase: "All the rater need to ratter any segment find" -- remove?
  • "# For instance, you can shift the window, and it will give you the same result" -- if no change is seen, then it's not a good example -- or am i misunderstanding?
  • ""is not a ChildRecordings class retrun null result"" -- should it just read ""is not a ChildRecordings class"? i'm guessing "return null result" was a message for us? or is it two error messages in one (ie EITHER no a class OR result is null)

internal things (low priority to these changes as they won't affect typical users)

• file_openner.R --> file_opener.R
• erro.convertion --> error.conversion
• add explanations & comments to data_to_OneHotEnc.R?
• convertor_long_cut.R is using a for loop, this could be done using a matrix & it would be a lot faster
• extractDataCR.R is using a for loop, this could be done using a matrix & it would be a lot faster
• find_rating_segment.R is using a for loop, this could be done using a matrix & it would be a lot faster
• hello.R is not used, right? remove?
• LENA.overlap.R uses a loop, replace with matrix; also, this is not how LENA is defined. LENA crosses out sections in which there is overlap with a 10ms stride, whereas this is nixing segments if there is any overlap
• in find_rating_segment.R what does this line mean: " time = seq(range_from-1000, range_to+1000, 1000)" -- does it mean that annotations that do not begin or end within one second of the window will not be discovered?
• find_rating_segment.R needs more comments, I personally cannot follow all the logic -- for instance, why "tmp$true_onset <- true_onset+1000"?

final thoughts

• we should make sure to clarify which packages & which versions of them we rely on - is that taken care of? what happens if user doesn't have right versions?
• is it a problem that you sometimes use DescTools:: and lubridate:: and other functions don't have the package named?

[alsonicr]

aggregate.rating OK reliability --> get.reliability SDT.raterData --> get.classification raterComparaison--> compare.rating

I found that in a R function the "." is use to creat a method for a class.

What do you think of :

• aggregateRating or aggregate
• getReliability
• getClassification
• compareRating

[alsonicr]

@alecristia I m pretty sure i change function name long ago it seem that some of the merging change it.

I'm going to check every comment you made to be sure

[alsonicr]

in find_rating_segment.R what does this line mean: " time = seq(range_from-1000, range_to+1000, 1000)" -- does it mean that annotations that do not begin or end within one second of the window will not be discovered? find_rating_segment.R needs more comments, I personally cannot follow all the logic -- for instance, why "tmp$true_onset <- true_onset+1000"?

@alecristia I will add comment on it .

First point : I need a larger time window because I can tell that on its two segments they can be defined as "without annotator". This can be useful for the diff function in line 140. If the annotations start at the first line then the diff will not notice the annotations.

Second point : This is because of the way diff works, the difference will be noted from the previous segment before the annotators are identified. So I have to identify the next segment to be correct.

[alsonicr]

harmonize how the class is referred to, eg ChildRecordings : a ChildRecordings class -- should it be ChildRecordingsData class?

I will say no since the ChildRecordings class do not really contain any data (except for metadata and integrity test). It may confuse childrecordsdata python user too maybe.

[alecristia]

any open items in this issue have been exported elsewhere